add plotter

sds4gdsp/plotter.py

@@ -0,0 +1,37 @@
+import os
+import shapely
+import geopandas as gpd
+import matplotlib.pyplot as plt
+from PIL import Image
+
+def get_route_fig(r):
+    fig, ax = plt.subplots(1, 1)
+    gpd.GeoSeries(r).plot(ax=ax, linewidth=5, zorder=1)
+    orig = shapely.geometry.Point([r.xy[0][0], r.xy[1][0]])
+    dest = shapely.geometry.Point([r.xy[0][-1], r.xy[1][-1]])
+    gpd.GeoSeries(orig).plot(ax=ax, color="red", markersize=250, zorder=2, alpha=0.8)
+    gpd.GeoSeries(dest).plot(ax=ax, color="green", markersize=250, zorder=2, alpha=0.8)
+    plt.axis("off")
+    ax.ticklabel_format(useOffset=False)
+    return fig
+
+def load_images(dirpath, extension=".jpg"):
+    imgs = []
+    for f in sorted(os.listdir(dirpath)):
+        if f.endswith(extension):
+            img_path = os.path.join(dirpath, f)
+            imgs.append(Image.open(img_path))
+    return imgs
+
+def plot_images(imgs, nrows=5, ncols=6, figsize=(10, 10)):
+    num_figs = nrows * ncols
+    num_imgs = len(imgs)
+    num_figs = min(num_figs, num_imgs)
+    fig, axes = plt.subplots(nrows=nrows, ncols=ncols, figsize=figsize)
+    for i, ax in enumerate(axes.flat):
+        if i < num_imgs:
+            ax.imshow(imgs[i])
+        ax.axis("off")
+    plt.tight_layout()
+    plt.show()
+    return fig

sds4gdsp/processor.py

@@ -16,6 +16,12 @@ def get_coords_from_graph(G: Graph, nodes: List[int]):
         coords.append((G.nodes[node]["x"], G.nodes[node]["y"]))
     return coords
 
+def convert_cel_to_point(cel_id: str, ref: pd.DataFrame):
+    """Convert cellsite to shapely point."""
+    coord = ref.loc[ref.uid==cel_id].coords.item()
+    point = loads(coord)
+    return point
+
 def dedupe_points(points: List[Point], distance_threshold: int):
     """Naive way to dedupe a points dataset given a distance threshold in meters."""
     pointpairs_df = pd.DataFrame(list(combinations(points, 2)), columns=["p1", "p2"])
@@ -50,8 +56,7 @@ def apply_softmax(arr: np.ndarray) -> np.ndarray:
 
 def convert_distances_to_probas(distances) -> np.ndarray:
     reversed_distances = distances ** -1
-    # probas = np.exp(-a*scaled_distances)
-    probas = apply_softmax(distances)
+    probas = apply_softmax(reversed_distances)
     return probas
 
 def scale_feature(feature, scaler) -> np.ndarray: